Implement StrCat in x86_64 assembly

Question

So here is the problem : I'm actually trying to recode some of the clib functions in assembly language (this is a school project to help start with assembly). The function I am currently working on is strcat. For the moment my goal is to keep it simple and follow the few following rules:

• If the destination string is NULL, return (in rax) the source string.
• If the source string is NULL, return (in rax) the destination string.
• Copy the source string at the end of the destination string (including terminating 0) and return (still in rax) the result.

Here is my code:

ft_strcat:
    push    rbp
    mov     rbp, rsp ; saving the stack state

    push    rdi ; seems to work better this way but I don't know why
    mov     rdi, [rsp + 24] ; destination string
    mov     rsi, [rsp + 16] ; source string

    push    rdi ; keeping the adress to return
    test    rsi, rsi ; in case one of the strings is NULL
    je      getdest
    test    rdi, rdi
    je      getsrc

    toend: ; to go to the end of the destination string
        cmp     byte [rdi], 0x0 ; is it the end?
        je      cpy ; if so, go to the next part
        inc     rdi ; else keep going
        jmp     toend ; loop

    cpy: ; to copy the source string to the end of the destination string
        mov     al, byte[rsi] ; getting the byte to copy
        mov     byte [rdi], al ; copying it
        cmp     byte [rsi], 0x0 ; it is the end of the source string?
        je      getdest ; if so, jump to the end
        inc     rdi ; else increase counter
        inc     rsi
        jmp     cpy ; loop

    getdest: ; if source is NULL or copy is done
        pop     rax
        jmp     end
    getsrc: ; if destination is NULL
        mov     rax, rsi
    end:

    pop     rdi ; get rdi back
    leave
    ret ; finally return...

I have tried a tremendous number of different ways (movsb, passing argument with the registers [directly], changing registers...) always reaching the same result :

• Segfault
• Strange characters in the string (if we can still call it a string...)

This current version keeps the destination part intact but adds those none-character characters at the end : Ph (this is just an example but the characters tend to change once in a while)...
I thought maybe you could help me (at least give me a hint of where to change things, or what may be wrong in my code), because I have looked all over the internet and never found things that would really help me.

Oh, by the way, I work with Nasm on Ubuntu (yes, I know ;) ).

Thx a lot to anyone who will answer. :)

Answer 1

most likely its

mov     rdi, [rsp + 32] ; destination string
mov     rsi, [rsp + 24] ; source string

or

mov     rdi, [rbp + 24] ; destination string
mov     rsi, [rbp + 16] ; source string

whichever you like better

In your current version, RSI contains return addr of the function. If currently you're appending some stuff to 'destination' then try

mov     rdi, [rsp + 24] 
mov     rsi, [rsp + 32]